Featured post

Indivisibility

Indivisible Movement

Indivisible Movement

The @IndivisibleTeam account was created on 18 December 2016 and I first wrote about it on 18 January 2017. Having poked around a bit, it would seem that this is going to be an ongoing thing for the next four years. This post will serve as an index of any work I do in this area.

The Indivisible Movement – visualizing the relationship between the four founder’s accounts and the main role account and the movement they have triggered.

The Indivisible Movement’s Conversation – visualizing the mentions originating from the 178 accounts identified in the first article.

The Indivisible Movement’s Facilitators – visualizing the accounts followed by 176 remaining accounts that were available on 20 January  2017, with an eye on identifying key human facilitators via identification of high centrality accounts.

Indivisible Influencers – visualized 254 movement accounts, who their influencers are with a focus on members of Congress, and then extracted their favorite topics, all seen from the perspective of the @Klout API.

If you are a reporter, a fellow blogger, or a student writing a paper and you’re interested in the graphics presented here feel free to drop a note to nrauhauser at gmail. Only a tiny fraction of the material I have ever makes it here and I occasionally take on producing visualizations for others.

Two Weeks Of Offensive Nonsense

Trump has been in office two weeks and he’s well on his way to offending pretty much everyone. This map by @BCAppelbaum shows where he has struck.

Trump Offenses

Trump Offenses

And domestically, Trump has made public statements about the “so called judge” in Washington that found Stephen Bannon’s poorly considered travel ban unconstitutional. Picking a fight with the judiciary is a surefire path to impeachment. Trump will not survive 2017 as president, he may not even make it to summer at the rate he is violating the principles that have held our country together for 240 years.

 

Muslim Ban? Fragile States?

 

Here’s Trump’s list of banned Muslim countries in red, and the ones where he has business interests are in gold. The unlabeled one at the uper right is Azerbaijan.

Trump's Muslim Ban Countries

Trump’s Muslim Ban Countries

And here’s a fragile states index for the region.

Fragile States Worst

Fragile States Worst

The banned countries are places where the governments have basically collapsed. People are complaining about the relationship between Trump businesses and the presence or absence of a ban. I’m not justifying, nor am I criticizing, I’m just noting that here is some data that hasn’t commonly appeared in conjunction with the coverage of the issue.

This map originally appeared in Fragile States Index 2016.

Trump Follower Infographic

Trump Follower Analytics

Trump Follower Analytics

I make my own graphs, but this one was too well done not to retain and feature here. The original by Rob Salkowitz is here: Trump’s 20 Million Twitter Followers Get Smaller Under The Microscope.

RogueNASA

Going Rogue

Responding to gag orders by President Trump, twenty three rogue accounts representing their respective agencies have been created on Twitter.

@ActualEPAFacts
@AltFedCyberz
@AltForestServ
@AltHHS
@AltNatParkSer
@AltUSFWS
@AltUS_ARC
@AltYosemiteNPS
@Alt_CDC
@Alt_NASA
@Alt_NIH
@AlternativeNWS
@BadHombreNPS
@ClimateNPS
@NatParkUndrgrnd
@NatlSciService
@RogueNASA
@RogueNOAA
@Rogue_USBGN
@altUSEPA
@alt_fda
@altusda
@ungaggedEPA
I decided to see who they were following. The Maltego graph was tolerable in terms of size, but these are best when I’m digging for details, not so helpful when I want to show others.

00-nps-maltego

My system already captures follower/following relationships and today I spent an hour crafting a script that would automate the production of an input file for Gephi. Those 23 accounts were following a total of 1,632 others and there were 2,433 total links.

01-nps-23-1632-2433

I wanted to see who was influential for these accounts so I squelched the low degree nodes to clean up the graph. The National Park Service is where the rebellion started so theirs is the oldest and most dense, clearly visible here. I was not at all surprised to find Bill Nye and Neil deGrasse Tyson prominent among those being followed.

02-nps-101-603-01

04-nps-101-603-04

03-nps-101-603-02

And it looks like @RogueNOAA is starting to join the fun.

05-nps-noaa

 

All in all, I don’t think Trump has any idea what he has unleashed here. The cute Twitter accounts are funny, but this level of conflict is going to embolden employees to say and do all sorts of things they would not in more settled time. First Manning, then Snowden, and who knows what might turn up next.

RogueNASA

Indivisible Influencers

The Indivisible Movement has expanded rapidly. I profiled 176 Twitter accounts in The Indivisible Movement’s Facilitators. Last night I started a study of who their influencers are last night and I found that my list now has 254 members – a 44% increase in 72 hours.

I have a set of @Klout transforms and I let them run on those 254 accounts, which led to this Maltego graph of 638 accounts total. The nodes on this graph are sized by the number of links, the one in the center is @IndivisibleTeam. These are the accounts that Klout finds to be influencers OF the Indivisibles.

01-ii-638-254-01

Within the 638 accounts on that graph I found fifty House members.

02-ii-house-01

02-ii-house-02

And fifty four Senators.

03-ii-senate-03

03-ii-senate-04

I’m not going to show the 260 other accounts that Klout found to be influential but I will make a few generalizations about them. Members of this group included:

  • Former presidents & presidential candidates.
  • Many news outlets.
  • Celebrities who’ve expressed opinions about Trump, mostly negative.
  • Activists who are in the middle of things but who do not have ‘Indivisible’ in their name.

I also let the transform run on topics for which the Indivisible accounts are influential. I tossed a lot of cruft – celebrities who are apolitical, keywords like ‘coffee’, that sort of thing.

04-ii-issues

I found four active categories of discussion

  • Issues topped the list – health care, civil rights, employment, and lastly war.
  • Some locations rose to prominence, but this didn’t make much sense to me given the broad geographic distribution of accounts.
  • WaPo and NBC seem to be favorite news sources.
  • Trump, Nixon, and Roosevelt were the presidents that were mentioned.

About the time I created the Indivisibility post, an index of my work in this area, I received a worried chat message.

Why is he trying to dox them?

The word ‘dox’ is a hacker/troll term, a verb that means “to identify the operator of an account, or to dig out a lot of personal information on an individual whose name is known”. The implication is that the person will then be harassed.

I don’t particularly care about who might be running any given Indivisible account. I’m currently taking Social and Economic Networks: Models and Analysis, I support what Indivisible and similar movements are trying to accomplish, and I’m just using this large social phenomenon as a testing ground for my (presumably) improved analysis capabilities.

I almost never approve comments on this blog. If you have a concern, leave a comment, and if it’s a well founded question I’ll probably post the text of it and a response.

The Resistance

I have been seeing #TheResistance hashtag in conjunction with The Indivisible Movement so I decided to take a closer look.

Using FollowerWonk revealed 4,800 profiles using the hashtag, so I took the somewhat unusual step of paying for a month of the service so I could download them all. I set a single account in the Netwar System to record them, and this morning I had 738 complete profiles.

When I combined the friends files for this group I found an amazing 378,00+ nodes with 1.16m links. This is what I’ve got after a half an hour of OpenOrd attempting to lay out the network. I dedicated seven cores to the job, my fan sounds like a jet engine, and the entire body of my MacBook Pro is warm.

01-openord

I sorted by degree and found there were 823 nodes with a degree of more than 100. Given the demands 15.3% of the data set have put on my system I’m curious to see if it will even import the whole thing.

It seems to be taking about 90 seconds per account to preserve them, so 4,800 accounts will take 7,200 minutes, or about five days. Once that’s done I’ve got an existing ELK server and a shiny new Kibi setup. I’m going to have a go at importing and indexing an estimated 15 million tweets. I don’t have a specific goal in mind with this, other than getting more familiar with what these two Elasticsearch systems can do.